Virtual testing of navigation and location of autonomous vehicles
Data about the precise location and speed are supplied by Global Navigation Satellite Systems (GNSS). With the help of additional correction methods, the position in three-dimensional space can be determined with an accuracy of a few decimetres. In addition, a variety of different sensors, including optical sensors such as lidar and radar, are used to detect obstacles and road markings. GNSS positioning and movement data can be exchanged with other vehicles (Car2Car) via wireless data connections or sent to a higher-level infrastructure (Car2X). In this way, potential accidents can be prevented early on, drastic braking manoeuvres are no longer necessary, and traffic jams are also becoming increasingly rare. With the aid of fleet control technology, it is even possible to identify vehicles that are not visible to the driver, for example because they are covered in curves by buildings or plants. The further development of ADAS systems and increasingly autonomous vehicles leads to greater safety and efficiency in road traffic.
To ensure the safety of ADAS and autonomous driving, millions of test kilometers must be driven on different roads in different environments. GNSS reception is not of constant quality. Especially in city centres and mountain areas, the reception of navigation data can be disturbed by the coverage of signals by buildings, bridges, vegetation or mountains. In addition, GNSS signals are reflected on flat and curved surfaces (multipath). In addition, there are additional sources of interference that can potentially block, interfere with or falsify GPS/GNSS receivers. Testing is therefore time-consuming and costly when all the required kilometres actually have to be driven.
In order to save costs and time and also increase safety during tests for involved and uninvolved persons, it is more efficient to realistically simulate the GPS/GNSS and interference signal environment and also the control of autonomous vehicles.
Spirent Communications plc, the English specialist for GNSS simulators, has developed such a simulation setup for the navigation and location of Connected Autonomous Vehicles (CAV). Simulators can be used to test autonomous vehicles or drones in the laboratory. The advantages are obvious: repeatable tests can also be carried out in realistically simulated inhabited environments, i.e. where the vehicles are ultimately to be used – and costs can also be saved.
A great deal of top-class technology is used. The basis is a dSPACE Scalexio computing system that was specially developed for HIL (Hardware-in-the-loop) projects. The satellite data is provided by a Spirent GNSS simulator. The vehicle simulation is performed by IPG CarMaker, a software for the virtual testing of passenger cars and light commercial vehicles. Microsoft AirSim, an open-source, cross-platform drone simulation software based on the “Unreal Engine” game engine, will be used for virtual testing of drones and swarms of drones. The test drives supported by navigation data are created with Spirent’s test scenario software SimGEN. The 3D simulation software SE-NAV from the French company Oktal-SE provides the final touch for a realistic signal representation. SE-NAV takes into account distortions of satellite signals by reflection, deflection and diffraction on surrounding surfaces. The cooperation between the GNSS SimGEN simulator software and the vehicle or drone simulator software is handled by Spirent’s SimREMOTE, an interface created for this purpose.
This simulation setup allows real-time testing of vehicles, aircraft or drones and analysis of their motion or trajectory, even in critical environments – without leaving the laboratory.
This video demonstrates the use of the simulator: https://youtu.be/BxSDuKfJirM
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
